Almost always have to revise or completely scrap AI outputs when I use it for content and design help for the podcast / newsletter

Very different experience from when I was using LLMs for a chrome extension side project I worked on

Using AI in subjective domains (e.g. content, design) only gets you ~90% of the way there. Still need a person with high taste to get real quality

Using AI in objectively verifiable domains (e.g. code, data formatting) makes it much easier to get high-quality results

If your writing is too wordy, no one will read it. If no one reads it, less value in writing it

If you can remove a word but preserve meaning, then remove it

Broke 10k subs on YouTube 🎉🎉

Thanks for all the support! It's been a lot of fun making these podcast videos and trying to improve the quality for the community

If you have any suggestions for future guests or the podcast in general I'm always open to feedback!

Principal Eng (IC8) from Meta interview footage is edited and ready for posting, will drop it tomorrow. We discussed:

• his promos from IC6 -> IC8
• why he accepts diffs that will break prod
• his notetaking system in vscode

Any guesses on who it is? 👀

Living in nyc is so much fun, but I think I'm spending ~2x what I was in sf if not more... combination of:

• More to do that I want to pay for
• Wanting to enjoy the moment more since being young in NYC feels temporary
• Things just being more expensive

Worth it though

"New grad hiring drops 50% compared to pre-pandemic levels", I doubt that this is fully due to AI tooling automating routine tasks though

Any ideas on the real reason behind this?

Chart below is from "The SignalFire State of Talent Report - 2025":

Interview with the Distinguished eng (IC9) ex-Meta, ex-OpenAI is ready! One of my favorite conversations so far, super excited to share with you all

Such interesting advice and stories about Zuck/John Carmack. Will drop the episode here tomorrow morning, follow if interested

New ergo office chair experiment, trying to find something better for my posture than the Herman Miller Aeron chair

Just interviewed an Principal Eng (IC8) from Meta, another one in the editing pipeline. We discussed:

• his promos at Meta from IC6 -> IC8
• his notetaking system
• IC8 impact working 40 hours a week

If interested, follow and I'll post about it here when it's ready

Just interviewed a Distinguished eng (IC9) ex-Meta, ex-OpenAI for the pod. One of my favorite conversations so far, super excited to share with you all

Such interesting advice and stories about Zuck/John Carmack. Will polish it up and launch it in a bit

If you do all the above, you'll increase your chances of getting a return offer dramatically. Good luck this summer!

4. Let people know - Don't work in a silo your whole internship. You're limiting yourself if your internship manager is the only person who can unblock you or knows about your work. Take the time to plug into the team and share updates.

3. Take initiative - If you did (1) and (2), you should have some spare time outside of your main project. Use that time to take on extra work or propose improvements on top of your original project.

2. Quality code - Don't just move fast though, your code needs to also be solid. You can measure this by how many revisions it takes for your code to land. You should aim for your code to land with 0 or 1 revisions by the 2nd half of your internship.

1. Move fast - Your project should be unambigious and not too complex. They're just trying to see if you can unblock yourself and make progress fast. If you can get your project done in half the internship's time that'd be excellent.

"How to get a return offer from FAANG internships" clip just went viral on my IG (~100k views, 907 shares, 1.7k saves)

I've had 5 interns at Meta, here's the gist on what a strong internship performance looks like:

Will start sharing more about the hardware side project if you want to follow along, have some stories about flying to China last minute and learning about how manufacturing works

Little known fact that I'm building a hardware side project. Was thinking through the Trump tariffs with my cofounder

We don't really have any other options aside from China. They do it faster and cheaper (even with the increased cost from tariffs)

Who is your favorite writer in tech/software engineering?

Looking for more quality examples to learn from

Had a lot of fun at the event and brought Ricky along too since he was UPE president back in the day

If any other college CS clubs are interested in hosting something similar DM me and maybe we can work something out one day

Did a talk at UCLA about "industry secrets I wish I knew before graduating college" for UPE & ACM clubs

If you're curious what we talked about, drop a reply and I'll share the recording link when it's ready:

Often the bottleneck for (2) is in how many people there are to delegate to and how hands-on you need to be in guiding them.

I do wonder what'll happen to the "coding machine" archetype of more senior engineers. Their outlier code output will be even crazier...

2. Resolving ambiguity, determining the most impactful direction, and scaling yourself through others - For this type of work, it'll become easier to delegate and scale yourself since you can pass more to AI rather than more junior engineers

Senior+ engineers understand that most scalable impact doesn't come from simple execution

Scalable impact comes from:

1. Writing high-leverage code that no one else can - Not something AI will replace in 12 months. I'm referring to specialists who do deep, domain-specific work

Anthropic CEO: "In 3-6 months, AI is writing 90% of the code. And then in 12 months we may be in a world where AI is writing essentially all of the code"

This scares junior engineers but not senior+ engineers. Here's why:

3. Communicate and gather feedback. In writing this is easy via comments. For speaking, watch if people zone out or not when you speak

I think about this more now that I enjoy creating content. When I think back on my career, doing this was a big part of the Staff promo at IG

They build great software, but fail to influence others.

3 fixes:

1. Learn the "taste" of your org. What is important? What do people care about? Why?

2. Watch what others who communicate well talk about

Biggest reason software engineers fail to get to Staff, through examples:

1. Give updates no one cares about
2. Present in meetings and everyone tunes out
3. Share opportunities that don't resonate

The fix for this problem:

Another great reason to take writing seriously as a software engineer.

About what tools to use: I just use simple docs (markdown, google doc, etc). You don't need a complex note taking system to benefit from writing.